Migraine affects over 6 million children and adolescents in the U.S. and leads to significant pain and disability. Cognitive behavioral therapy (CBT) effectively reduces the number of headache days in pediatric patients. CBT includes distinct components, including mind and body approaches (deep breathing, progressive muscle relaxation, imagery, biofeedback) and cognitive reappraisal training (CR). While the new strategic plan of the National Center for Complementary and Integrated Health highlights ?Neurobiological Effects and Mechanisms? as a top scientific priority for non-pharmacological pain research, the brain mechanisms by which components of CBT work remain poorly delineated. To dissect the neural mechanisms engaged by CBT, we will conduct a mechanistic clinical trial to compare a sample of 80 youth (ages 10 to 17) with migraine who receive relaxation training (BART) to 80 who receive CR training. Functional MRI and quantitative sensory testing (QST) data will be obtained before and after intervention (an 8 week period) to address these aims: Aim 1A. Do BART and CR training differentially alter resting brain activation in a fashion related to headache reduction? Our pilot data indicate migraine patients have pro-nociceptive patterns of functional connectivity with the amygdala while CBT produces anti-nociceptive patterns of amygdalar functional connectivity. We hypothesize that BART will preferentially activate the ventromedial prefrontal cortex (vmPFC) while CR training will preferentially activate the dorsolateral prefrontal cortex (dlPFC) and other regions of cognitive control. To test this, arterial spin labeling (ASL) MRI will be used to compare pre vs. post training differences in steady state brain activation in BART and CR groups. Resting state BOLD fMRI will be used to determine if BART and CR training produce differential alterations of the functional connectivity of the vmPFC and dlPFC with the amygdala. Aim 1B. Does active performance of BART and CR differentially recruit activation of the vmPFC and dlPFC? Preliminary data from CBT were obtained during rest and may reflect tonically active consequences of therapy/headache reduction rather than the phasic mechanisms involved during active engagement of BART/CR strategies. To further determine if BART and CR involve distinct brain mechanisms, both ASL and BOLD MRI will be used to compare brain activation and amygdalar functional connectivity during active performance of each treatment vs. rest. Aim 2. Does the efficiency of pain modulation predict headache reduction following BART vs. CR? QST assessment of the efficiency of conditioned pain modulation (CPM) provides an index of the involvement of endogenous pain modulatory mechanisms in both pain conditions and pain treatments. Our preliminary work has shown that low efficiency of CPM is predictive of headache reduction following CBT. However, it remains unknown if BART vs. CR is more effective in the presence of disrupted endogenous pain modulation. Thus, pre-treatment CPM will be used to test the hypothesis that low efficiency of CPM will be differentially predictive of headache reductions following BART vs. CR.